I started a rumor on TikTok and Twitter.

In less than 90 minutes, news sites were reporting it as fact.

Donald Trump’s staff assaulted an employee at Arlington National Cemetery and broke multiple laws while using the site for campaigning. They desecrated graves when their fleet of SUV’s drove and parked over them. Donald Trump posed next to the family of the deceased smiling and giving a thumbs-up.

That wasn’t the plan, though. That wasn’t what they wanted the story to be about.

The plan was to falsely claim that Vice President Kamala Harris skipped a public event at Arlington as a sign of disrespect to the fallen soldiers.

There was no public event and campaigning (and even most photography) is banned at Arlington.

If not for the disturbing events taking place that day, Trump’s team would have successfully spread a lie meant to harm Harris (though they still tried anyway, with little success).

The far-right creates false stories and uses propaganda sites to back it up all of the time – there are entire Twitter accounts dedicated just to doing that to me (yes, four years later).

The organized disinformation machine is somewhat synchronous – coordinated efforts to promote a lie they know is a lie by intentionally manipulating the algorithms.

But that’s not the only kind of rumor that spreads.

Then there are accidental rumors, too, like the joke about JD Vance’s sofa-sexual relations, started in jest by a relatively unknown Twitter user.

The Vance joke was obviously always that – a joke. 

What if a less-obviously false rumor targeting the GOP took off? Could it manipulate the algorithm and media to propagate the lie? What would the data say about the speed and nature of a rumor if you had full access to the account data of the person starting the rumor?

The difficulty in measuring how a rumor or disinformation spreads online is that we’re always outside observers. Few with a platform as large as mine are willing to engage in social experiments of this nature because of the obvious ethical questions and because they don’t want to lose followers.

As someone unafraid to lose followers and well-versed in the mechanisms social media uses to spread disinformation, I started a rumor and carefully monitored its reach.

The rumor: Melania Trump had filed for divorce.

I picked something rather innocuous – Trump’s already been divorced multiple times and he was convicted of 34 felony counts earlier this year for trying to hide his affair with Stormy Daniels. It wouldn’t surprise anyone, and I doubt it would have much of an impact overall in the election if she did file for divorce.

I provided almost no details. I even named the wrong Florida county initially (they live in Palm Beach, I said Miami-Dade). I made up a random time (5:12 pm) and said the filing was made through the e-file court system, which allows anyone to file any civil action 24 hours, 7 days a week. 

Because the court systems take days to reflect these filings, there was no easy way to prove the story was false absent a public statement by Melania herself. I posted something that would require a direct statement from the subject in order to disprove. It could not be fact-checked.